DeepSeek-R1 Base: pricing, context window and benchmarks
DeepSeek-R1 Base is a multimodal model from DeepSeek. It offers a 400,000-token context window, costs $5.05 per million input tokens and $26.24 per million output tokens, and averages 80.1 across the index's benchmark suite.
Pricing
DeepSeek-R1 Base costs $5.05 per million input tokens and $26.24 per million output tokens. A blended 3:1 input-to-output mix works out to about $10.35 per million tokens. Cached input is listed at $0.59 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $5.05 | per 1M tokens |
| Output | $26.24 | per 1M tokens |
| Cached input | $0.59 | per 1M tokens |
| Blended 3:1 | $10.35 | estimate per 1M tokens |
| 1M in + 1M out | $31.29 | estimate |
Benchmarks
Scores below are modelled estimates on a 0–100 scale. Use them to compare relative position, not as official vendor results.
| Benchmark | Score | Rank | Percentile | Relative |
|---|---|---|---|---|
| MMLU-Pro | 81.0 | 147 | 76th | |
| GPQA Diamond | 76.2 | 114 | 81th | |
| SWE-bench Verified | 71.2 | 91 | 85th | |
| HumanEval | 83.4 | 149 | 75th | |
| MATH-500 | 84.9 | 109 | 82th | |
| AIME 2025 | 73.8 | 189 | 69th | |
| LiveCodeBench | 76.0 | 82 | 86th | |
| IFEval | 88.2 | 97 | 84th | |
| TruthfulQA | 82.2 | 126 | 79th | |
| HellaSwag | 90.0 | 85 | 86th | |
| MMMU | 74.2 | 178 | 70th | |
| Arena Elo | 1442.9 | 94 | 84th |
What DeepSeek-R1 Base is good for
DeepSeek-R1 Base is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $5.05 per million input tokens, it sits in the mid-range of the index, and the 400,000-token context window is enough for large files and multi-file diffs.
Strengths
- Large 400,000-token context window for long documents and agent traces.
- Open weights make self-hosting, fine-tuning and data-residency review possible.
- Accepts more than text, including image input where the provider supports it.
- Coding benchmarks are a relative strength within its tier.
Trade-offs
- No standout weakness in the indexed specifications, but real-world behaviour still needs testing.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| GLM-4.6 | Zhipu AI | 128000 | $5.10 | 80.2 |
| Kimi K1.5 Chat | Moonshot AI | 128000 | $5.11 | 80.2 |
| MiniMax-M2 Chat | MiniMax | 200000 | $5.02 | 80.2 |
| DeepSeek-R1 Base | DeepSeek | 400000 | $5.05 | 80.1 |
| Claude Haiku 4.5 Turbo | Anthropic | 200000 | $14.37 | 80.0 |
| ERNIE 4.5 | Baidu | 65536 | $13.44 | 80.0 |
FAQ
How much does DeepSeek-R1 Base cost?
DeepSeek-R1 Base is listed at $5.05 per million input tokens and $26.24 per million output tokens. Cached input is $0.59 per million tokens. A blended 3:1 input-to-output mix is about $10.35 per million tokens.
What is the context window of DeepSeek-R1 Base?
DeepSeek-R1 Base has a context window of 400,000 tokens and can emit up to 65,536 output tokens per request in this index.
Is DeepSeek-R1 Base open weights?
Yes. DeepSeek-R1 Base is listed as an open-weight model under a open weights licence.
How does DeepSeek-R1 Base score on benchmarks?
DeepSeek-R1 Base averages 80.1 across the scored benchmarks in this index, ranking #118 of 600 models. Its strongest relative result is Arena Elo at 1442.9.
What is DeepSeek-R1 Base best for?
DeepSeek-R1 Base is tuned toward coding work. It is a reasonable candidate for code generation, review, debugging and repository-scale tasks, especially when paired with a retrieval layer. At $5.05 per million input tokens, it sits in the mid-range of the index, and the 400,000-token context window is enough for large files and multi-file diffs.